Re: [NGO] external module properties

[Andy Bierman <ietf at andybierman.com>]

David Harrington wrote:
>  
> 
>> We already have a single naming scope for module names.
>> There is no concept at all in YANG of the Acme FOO-MIB
>> and some other FOO-MIB.  There is just one FOO-MIB,
>> and any import (or include) which identifies FOO-MIB
>> in the target is referring to the one and only possible
>> module (within that implementation) called FOO-MIB.
> 
> This works within an implementation. How does it work in an NMS that
> deals with FOO-MIB instances in different implementations? If both
> Ciso and Nortel publish a FOO-MIB, how does an NMS managing both
> vendors' devices differentiate which FOO-MIB specification to use for
> each managed node, especially if the BAR-MIB imports "FOO-MIB"?
> 

We already have this problem now with SMIv2 module names.
The standard uses 'stock fuzzy text' to warn vendors
to pick module names that will not clash.  This is how the IETF
punts the problem, because the real solution involves a registry,
and we apparently already have enough of those.  Vendors would
not want to register their module names with IANA anyway.


> David Harrington

Andy

_______________________________________________
NGO mailing list
NGO at ietf.org
https://www.ietf.org/mailman/listinfo/ngo